CN1429433A - Multiuser Detection Using Adaptive Combination of Joint Detection and Serial Interference Cancellation - Google Patents

Abstract

A time division duplex communication system transmits a plurality of data signals using the same shared spectrum in a time slot using code division multiple access. A combined signal may also be received over the shared spectrum of the time slot. The plurality of data signals are divided into a plurality of data signal groups. The combined signal is matched filtered based on a partial symbol response associated with the data signal of one of the groups. Data of each data signal in the same group is detected simultaneously. An interference signal is generated based in part on the detected data for the group. The generated interference signal is subtracted from the combined signal. The data of the other groups is detected by processing the subtracted signal.

Description

Multiuser Detection Using Adaptive Combination of Joint Detection and Serial Interference Cancellation

This application claims U.S. Provisional Patent Application No. 60/189,680, filed March 15, 2000, and U.S. Provisional Patent Application No. 60, filed May 26, 2000. .60/207,700 priority.

Background of the invention

The present invention generally relates to wireless communication systems. In particular, the present invention relates to joint detection of multi-user signals in a wireless communication system.

FIG. 1 is an illustration of a wireless communication system 10 . The communication system 10 has base stations 12 ₁ to 12 ₅ communicating with user equipments (UEs) 14 ₁ to 14 ₃ . Each base station 12 ₁ has a corresponding working area, and in the working area, the base station communicates with the user equipments 14 ₁ to 14 ₃ in the area.

In some communication systems, such as code division multiple access (CDMA) and time division duplex system (TDD/CDMA) using code division multiple access, multiple communications are sent using the same frequency spectrum, and each communication is generally transmitted through its signal chip. Code sequences are distinguished. In order to use the frequency spectrum more effectively, the TDD/CDMA communication system uses repeated frames divided into several time slots for communication. According to the different bandwidth of the communication, the communication sent in this type of system has one or more associated signal chip codes and time slots.

Because such systems can send multiple communications simultaneously using the same frequency spectrum, receivers in such systems must distinguish between the communications. One way to detect such signals is matched filtering. Matched filtering detects communications sent with a single code while treating other communications as interference. Therefore, to detect multiple codes, a corresponding number of matched filters must be used. Another method is Serial Interference Cancellation (Successive InterferenceCancellation SIC). The method detects one channel of communication, and then subtracts the signal base value of the channel of communication from the received signal to detect the next channel of communication.

In some applications, in order to improve the communication performance, it is required to be able to detect multiple communication at the same time. Simultaneous detection of multiple communications is called joint detection. Some joint detectors use Cholesky decomposition for minimum mean square error (MMSE) detection and use a zero-forcing block equalizer (ZF-BLE). These detectors are complex and require extensive receiver resources.

Therefore, it is necessary to find alternative methods for multi-user detection.

Contents of the invention

A time-division duplex communication system that uses a code-division multiple access method to transmit multiple data signals using a shared frequency spectrum in one time slot. A combined signal is received through the shared spectrum of the time slot. The multiple data signals are divided into multiple data signal groups. The combined signal is matched filtered based on the partial symbol response associated with the data signal of one of the groups. The data of each data signal in the same group is detected simultaneously. The interference signal is generated based on the partial detection data of the group. The interfering signal is subtracted from the combined signal. By processing the subtracted signal, other sets of data can be detected.

Description of drawings

Figure 1 shows a wireless communication system.

Figure 2 shows a simplified transmitter and receiver using joint detection

Figure 3 is a diagram of a communication burst.

Figure 4 is a flowchart of the adaptive combination of joint detection and serial interference cancellation

Figure 5 is a diagram of a joint detection and serial interference cancellation adaptive combination device

Figure 6-12 is a performance comparison chart of joint detection and serial interference cancellation adaptive combination, complete joint detection and RAKE receiver.

Detailed ways

2 is a diagram of a simplified transmitter 26 and receiver 28 for a joint detection (JD) and serial interference cancellation (SIC) adaptive combination "SIC-JD" used in a TDD/CDMA communication system. In _a typical system, a transmitter 26 is located in each UE _141-143 , and a plurality _{of transmit circuits 26 for transmitting multiplex communications are located in each base station 121-125 .} The base station 12 ₁ requires each effective communication UE 14 ₁ to 14 ₃ to have at least one transmitting circuit 26 . The SIC-JD receiver 28 may be located in the base station ₁₂₁ , in the UEs ₁₄₁ to ₁₄₃ , or both. The SIC-JD receiver 28 receives communications from the plurality of transmitters 26 or transmission circuits 26 .

Each transmitter 26 transmits data over a radio channel 30 . The data generated by the data generator 32 in the transmitter 26 is transmitted to the receiver 28 via a reference channel. Depending on communication bandwidth requirements, reference data is assigned to one or more codes and/or time slots. The modulating and spreading device 34 spreads the reference data, and uses the training sequence in the specified time slot or code to change the extended reference data into time-division multiplexed data. The resulting sequence is called a communication burst. The communication burst is modulated to radio frequency by modulator 36 . Antenna 38 radiates radio frequency signals via radio channel 30 to antenna 40 of receiver 28 . The modulation type used for such transmission communication may be any type known to those skilled in the art, such as direct phase shift keying (DPSK) or quadrature phase shift keying (QPSK).

As shown in FIG. 3, a typical communication burst 16 has a training sequence 20, a guard period 18 and two data bursts 22,24. The training sequence 20 separates the data bursts 22, 24, and the guard period 18 separates the communication bursts, so that bursts from different transmitters arrive at different times. The two data bursts 22, 24 contain the data of the communication burst and generally have the same symbol length. The midamble contains a training sequence.

Antenna 40 of receiver 28 receives various radio frequency signals. The received signal is demodulated by the demodulator 42 to generate a baseband signal. The baseband signals are processed by, for example, channel estimation means 44 and SIC-JD means 46 in time slots and using corresponding communication burst codes assigned to corresponding transmitters 26 . The channel estimation means 44 provides channel information, such as the channel impulse response, using the midamble components in the baseband signal. The SIC-JD means 46 then uses the channel information to estimate the transmission data of the received communication burst as hard symbols.

The SIC-JD means 46 utilizes the channel information provided by the channel estimating means 44 and the known spreading code used by the transmitter 26 to estimate various received communication burst data. Although the SIC-JD device 46 is described herein in conjunction with a TDD/CDMA communication system, the method is also applicable to other communication systems, such as CDMA.

Figure 4 illustrates a method of performing SIC-JD in a specific time slot in a TDD/CDMA communication system. In this specific time slot, several communication bursts are superimposed on each other, for example K communication bursts. The K bursts can come from K different transmitters. However when some transmitters use multiple codes in that particular time slot, then the K bursts can come from fewer than K transmitters.

Both data bursts 22, 24 in the communication burst 16 have a predetermined number of transmitted symbols, eg _Ns . Each symbol is transmitted using a predetermined number of chips of a spreading code, which is the spreading factor (SF). In a typical TDD communication system, the communication data of each base station (12 ₁ to 12 ₅ ) contains a related scrambling code, and the scrambling code can distinguish the base stations from each other. In general, scrambling codes do not affect the spreading factor. Although the terms "spreading code" and "spreading factor" are still used hereinafter for systems using scrambling codes, for the following cases, "spreading code" will refer to the combination of the scrambling code and the spreading code. Thus, the data bursts 22, 24 each have _Ns x SF signal slices. After passing through a channel with W signal slice impulse responses, the length of each received burst is SF×N _s +W-1, which can also be represented by Nc signal slices. The code of the K ^th pulse train in the K pulse trains is denoted by C ^(k) .

Each K ^th burst is received by the receiver and can be expressed by the following equation 1:

r ^(k) = A ^(k) d ^(k) , k = 1...K Formula 1

r ^(k) is the received base value of the K ^th burst, and A ^(k) is the combined channel response (a N _c ×N _s matrix). The j ^th column in the matrix A ^(k) is the zero-filled form of the j ^th element of d ^(k) in response to s ^(k) . The symbol response s ^(k) is the convolution of the estimated response h ^(k) of the burst with the spreading code C ^(k) . d ^(k) is the unknown data symbol sent in the burst. The estimated response h ^(k) of each K ^th pulse train has a length of W chips and can be expressed by the following Equation 2: ${\underset{\‾}{h}}^{\left(k\right)}={\mathrm{\γ}}^{\left(k\right)}\&Center\; Dot;{\underset{\‾}{\overset{~}{h}}}^{\left(k\right)}$ Formula 2

表示组群特有的信道响应。对于上行线路通信，各脉冲串的 h ^(k)、γ^(k)以及 互不相同；对于下行线路通信，各脉冲串的

均相同，而γ^(k)不同。而如果在下行线路中采用传输分集制，则各脉冲串的γ^(k)和 均互不相同。where γ ^(k) represents transmitter gain and/or path loss; represents the burst-specific fading channel response; while for a burst group of similar channels,

Indicates the group-specific channel response. For uplink communication, h ^(k) , γ ^(k) and are different from each other; for downlink communication, the

are the same, but γ ^(k) is different. However, if transmit diversity is adopted in the downlink, the γ ^(k) and are different from each other.

The total vector of all K bursts received over the radio channel can be expressed by Equation 3: $\underset{\‾}{r}=\underset{i=1}{\overset{k}{\mathrm{\Σ}}}{\underset{\‾}{r}}^{\left(k\right)}+\underset{\‾}{\mathrm{no}}$ Formula 3

where n represents a zero-average noise vector.

If A ^(k) of all data bursts is combined into matrix A, and all unknown data d ^(k) of each burst is combined into matrix d , then Equation 1 becomes Equation 4.

r = A d + n Formula 4

The received power of each K ^th burst is determined by the SIC-JD based on a priori knowledge of the receiver 28, burst channel estimation from the burst specific training sequence, or a matched filter bank. The K bursts are arranged in descending order according to their measured received power.

Pulse trains of substantially the same power level (eg, within the same threshold) are grouped together and arranged in G groups, 48 . The G groups are arranged in descending order according to the power of each group, for example, the order is from group 1 to G, and the group 1 has the highest received power. FIG. 5 is an illustration of SIC-JD performed by the SIC-JD device 46 according to G groups.

For group 1 with the highest received power, only the burst symbol response matrix _Ag ⁽¹⁾ in this group is determined, which matrix contains only the symbol responses of the bursts in group 1. At the same time the reception vector r of group 1 is denoted by x _g ⁽¹⁾ . Thus, for group 1, Equation 4 becomes Equation 5.

x _g ⁽¹⁾ = A _g ⁽¹⁾ d _g ⁽¹⁾ + n Formula 5

Where d _g ⁽¹⁾ is the data in the group 1 burst. Equation 5 embodies the effects of intersymbol interference (ISI) and multiple access interference (MAI). Therefore, the effects of the other groups (Group 2 to Group G) were not considered.

After the received vector x _g ⁽¹⁾ is filtered by the matched filter 66 ₁ of group 1, it becomes the symbol response of the burst in group 1, and the process is expressed by formula 6, 50. ${\underset{\‾}{\mathrm{the\; y}}}_g^{\left(1\right)}=A_g^{\left(1\right)h}{\underset{\‾}{x}}_g^{\left(1\right)}$ Formula 6

Among them, y _g ⁽¹⁾ is the result of matched filtering.

The joint detection device 68 ₁ of group 1 performs joint detection on group 1, and uses the matched filtering result y _g ⁽¹⁾ to make soft decision estimation . A joint detection method is to calculate the least squares and zero-forcing results according to Formula 7. ${\underset{\‾}{\overset{^}{d}}}_{g.\mathrm{soft}}^{\left(1\right)}={\left(A_g^{{\left(1\right)}^h}{A}_g^{\left(1\right)}\right)}^{-1}{\underset{\‾}{\mathrm{the\; y}}}_g^{\left(1\right)}$ Formula 7

for Hermitian form. Another way is to calculate the minimum mean square error result according to Equation 8. ${\underset{\‾}{\overset{^}{d}}}_{g.\mathrm{soft}}^{\left(1\right)}={(A_g^{{\left(1\right)}^h}{A}_g^{\left(1\right)}+{\mathrm{\σ}}^{2}I)}^{-1}{\underset{\‾}{\mathrm{the\; y}}}_g^{\left(1\right)}$ Formula 8

where I is the identity matrix and ^σ2 is the standard deviation.

与 均为条状码组托普利兹矩阵，因此公式7或8求解过程的复杂度也得到了降低。另外，在运用Cholesky分解时所导致的性能降低也可忽略不计。对多个的脉冲串进行Cholesky分解是相当复杂的，然而当对一个较小的用户组进行Cholesky分解时，其复杂度可显著降低。This method only jointly detects a group of pulse trains, which has the advantage that the complexity of analyzing a single group is lower than that of analyzing all signals. because

and Both are barcode group Toeplitz matrices, so the complexity of the solution process of formula 7 or 8 is also reduced. In addition, the performance degradation caused by applying Cholesky decomposition is also negligible. Performing Cholesky decomposition on multiple bursts is quite complex, but when performing Cholesky decomposition on a smaller user group, the complexity can be significantly reduced.

soft verdict - hard verdict block 70 ₁ will soft verdict Received data as group 1 is converted to hard decision , 54. When processing other low power groups, the multiple access interference generated by group 1 to the low power group is estimated by a group 1 interference generation block 72 ₁ according to Equation 9, 56 . ${\underset{\‾}{\overset{^}{r}}}^{\left(1\right)}=A_g^{\left(1\right)}{\underset{\‾}{\overset{^}{d}}}_{g.\mathrm{hard}}^{\left(1\right)}$ Formula 9

为组1赋予 r的估算基值。in

Assign an estimated base value for r to group 1.

For the adjacent group 2, the estimated base value of group 1 is subtracted (for example by subtractor 74 ₁ ) from the received vector x _g ⁽¹ ) to obtain x _g ⁽²⁾ , as shown in equation 10,58. ${\underset{\‾}{x}}_g^{\left(2\right)}={\underset{\‾}{x}}_g^{\left(1\right)}-{\underset{\‾}{\overset{^}{r}}}^{\left(1\right)}$ Formula 10

将从组2的干扰取消信号中减掉(例如通过减法器24₂)，从而得到 ${\underset{\‾}{x}}_g^{\left(2\right)}-{\underset{\‾}{\overset{^}{r}}}^{\left(2\right)}={\underset{\‾}{x}}_g^{\left(3\right)}$ ，62。使用该程序，可以逐次对各组进行处理，直至最后的组G。由于组G为最后一组，因此不需要生成干扰信号。从而组G仅需使用组G匹配滤波器66_G、组GJD块68_G以及用于恢复硬字符软判决—硬判决块70_G，64。As a result, multiple access interference generated by group 1 in the received signal can be effectively eliminated. The next strongest power group (i.e. group 2) uses x _g ⁽²⁾ and passes through the group 2 matched filter 66 ₂ , the group 2 JD block 68 ₂ , the soft-hard decision block 70 ₂ and the group 2 interference generation block 72 ₂ , 60 for similar processing. The generated Group 2 interference

will be subtracted (e.g. by subtractor 24 ₂ ) from the interference canceling signal of group 2 to obtain ${\underset{\‾}{x}}_g^{\left(2\right)}-{\underset{\‾}{\overset{^}{r}}}^{\left(2\right)}={\underset{\‾}{x}}_g^{\left(3\right)}$ , 62. Using this procedure, each group can be processed successively, up to the last group G. Since group G is the last group, no interference signal needs to be generated. Group G thus only needs to use the group G matched filter 66 _G , the group GJD block 68 _G and the soft decision-hard decision blocks 70 _G , 64 for recovering hard characters.

When performing SIC-JD on _UE141 , it is not necessarily necessary to process all groups. If all the bursts to be received by UE14 ₁ are in the group with the highest received power or higher received power, UE14 ₁ only needs to process the group containing its bursts. Therefore, the processing required by UE ₁₄₁ can be further simplified. The simplification of UE14 ₁ processing reduces power consumption, thus extending battery life.

Since the N _c ×K·N dimensional matrix is replaced by G JD dimensional levels N _c ×n _i ·N _s (where i=1 to G, and n _i is the number of bursts in the i ^th group), the SIC-JD The complexity is lower than single-step JD. The complexity of JD scales from the square to the cube of the number of bursts to be jointly detected.

The advantage of this method is that it achieves a balance between computational complexity and performance. If all the bursts are placed in a single group, the solution problem can be reduced to a JD problem. Single grouping can be achieved by forcing all bursts into the same group, or by using a wider threshold. On the other hand, when the group contains only one signal or only one signal is received, the solution can be reduced to a SIC-LSE problem. Instead, this condition can be achieved using a narrower threshold, or by hard-limiting the size of each group to force each burst into its own group. By choosing the threshold, the balance between performance and computational complexity can be arbitrarily achieved.

Figures 6 to 12 show the simulation results of the bit error rate (BER) performance comparison between SIC-JD and full JD and RAKE receivers under various multipath fading channel conditions. The selected parameters are 3G UTRA TDD CDMA system parameters: SF=61; W=57. Each TDD burst/slot is 2560 chips or 667 microseconds long. These bursts have two data fields each with N _s QPSK symbols, a training sequence field and a guard period. Each simulation runs over 1000 time slots. In any case, the number K of bursts is chosen to be eight. It is assumed here that all receivers know exactly the channel response of each burst so that they can be sequenced and grouped correctly. It is also assumed that the channel response is a time-invariant response over one slot, while successive slots are subject to non-correlated channel responses. Channel coding was not used in this simulation. The JD algorithm jointly detects all K bursts. RAKE-type receivers use a matched filter bank for an i ^th burst code ( ${\underset{\‾}{\overset{^}{d}}}^{\left(i\right)}=A^{{\left(i\right)}^h}{\underset{\‾}{r}}^{\left(i\right)}$ ). Maximum Ratio Combiner (MRC) stages are implicit in these filters because they match the overall symbol response.

The performance simulation is carried out using the multipath program file defined by the ITU channel model under fading channel conditions. The ITU channel model includes Indoor A, Pedestrian A, Vehicular A models, and 3GPP UTRA TDD Working Group 4 Case 1, Case 2 and Case 3 Model. In Vehicular A and Case 2 channels, SIC-JD experienced a maximum 1 decibel (dB) degradation compared to full JD in the range of 1%-10% BER. For all other channels, the performance of SIC-JD deviates from full JD within 0.5dB. Since Vehicular ACase 2 is the worst case of all the cases studied, only its performance curve is presented. Among all channels simulated, Vehicular A and Case 2 channels have the largest delay spread. Vehicular A is a 6-branch model with relative delays of 0, 310, 710, 1090, 1730 and 2510 nanoseconds, and relative average powers of 0, -1, -9, -10, -15 and -20 decibels (dB ). Case 2 is a 3-branch model, each branch has the same average power and relative delays are 0, 976 and 1200 nanoseconds, respectively.

，i＝1至K)且均相同，而γ⁽ⁱ⁾(i＝1至K)不同。在δ⁽ⁱ⁾的选择上，使按照功率级别布置脉冲串时，两个相邻的脉冲串之间有一个2dB的功率差。例如，此种功率差会存在于基站12₁针对不同的UE(14₁-14₃)脉冲串施加不同传输增益的下行线路中。图6和图7表明，在1％-10％的误码率(BER)范围内，与JD相比，SIC-LSE经受的递降不大于1dB。这正是通常人们所关心的非编码BER(原始BER)范围。由于不能优化处理ISI，RAKE接收机出现了显著递降。由于脉冲串之间功率差增大，SIC-LSE的性能得到了提高，且当功率差为1-2dB(取决于不同的信道)时，其性能即可与完全JD相媲美。Figures 6 and 7 compare the bit error rate (BER) and chip-level signal-to-noise ratio (SNR) performance of the SIC-LSE receiver with full JD and RAKE-like receivers under two multipath fading channel conditions. The group size is mandatory set to 1 to form K groups in both sender and receiver. Also shown is the theoretical value of the binary phase-shift keying (BPSK) bit error rate (BER) in an additive white Gaussian noise (AWGN) channel; the AWGN channel specifies a lower bound for the BER. BER is averaged over all bursts. Figure 6 shows an example of different channels, where it is assumed that the fading channels through which the bursts pass are independent, but that all channels have the same average power resulting in the same average signal-to-noise ratio (SNR). In this case, (i=1 to K) are different, and γ ⁽ⁱ⁾ (i=1 to K) are all the same. This situation exists in the uplink where power control only compensates for long-term fading and/or path loss but not for short-term fading. In each time slot, the bursts are arranged in power according to the corresponding h ⁽ⁱ⁾ (i=1 to K). Figure 7 shows a similar graph for the common channel case. In this figure it is assumed that all bursts pass through the same multipath channel (i.e.

, i=1 to K) and are the same, but γ ⁽ⁱ⁾ (i=1 to K) are different. In the selection of δ ⁽ⁱ⁾ , when the pulse trains are arranged according to the power level, there is a 2dB power difference between two adjacent pulse trains. For example, such a power difference exists in the downlink where the base station 12 ₁ applies different transmission gains to different UE (14 ₁ -14 ₃ ) bursts. Figures 6 and 7 show that in the bit error rate (BER) range of 1%-10%, SIC-LSE suffers no more than 1 dB of degradation compared to JD. This is the non-coded BER (raw BER) range that people are usually concerned with. RAKE receivers suffer significant degradation due to inability to optimally handle ISI. Due to the increased power difference between bursts, the performance of SIC-LSE is improved, and when the power difference is 1-2dB (depending on different channels), its performance is comparable to that of full JD.

(i＝1至n_s)相同。其中n_s为g^th组中脉冲串数。这潜在地代表了上行线路中的多代码情况，在该情况中，每一个UE14₁发送2个代码。SIC-JD接收机28将与同一个UE14₁相关的多个代码编在同一个组内，从而形成4个组。图10和图11展示了公共信道的情形。此图中假定所有脉冲串组均通过同一条多径信道，即 (g＝1至n_s)均相同，而γ_g(g＝1至G)不同。选择γ_g时，使依据功率级别安排脉冲串组时，两个相邻的组之间有一个2dB的功率差。这潜在地代表了下行链路中的多代码情况，在此情况下，基站12₁为每一个UE14₁发送2个代码。图10和图11所示趋势与图8和图9中所示观察到的SIC-LSE的性能趋势相似。在1％-10％BER范围内，SIC-LSE的性能与JD相当(即差别不大于1dB)，而该范围正是人们所关心的非编码BER的工作范围。当功率差为1-2dB(取决于不同的信道)时，SIC-LSE的性能即可与完全JD相媲美。如图所示，其性能随两个脉冲串之间功率差的增大而提高。Figures 8, 9, 10 and 11 compare the BER and SNR performance of the SIC-JD receiver with full JD and RAKE class receivers under two multipath fading channel conditions. Each of the 8 codes is divided into 4 groups of two codes in the transmitter and receiver. The BER is averaged over all bursts. Figures 8 and 9 show examples of distinct channels, in which it is assumed that the fading channels through which the burst groups pass are independent, but that all channels have the same average power resulting in the same average SNR. All bursts in the same group get the same channel response. In this situation, (g=1 to G) are different from each other, and the channel response of each burst in the group

(i=1 to n _s ) are the same. Where n _s is the number of bursts in the g ^th group. This potentially represents a multi-code situation in the uplink, where each UE 14 ₁ sends 2 codes. The SIC-JD receiver 28 compiles multiple codes related to the same UE 14 ₁ into the same group, thereby forming 4 groups. Figure 10 and Figure 11 show the common channel situation. In this figure, it is assumed that all burst groups pass through the same multipath channel, that is, (g=1 to n _s ) are all the same, but γ _g (g=1 to G) are different. When γ _g is selected, when the burst groups are arranged according to the power level, there is a 2dB power difference between two adjacent groups. This potentially represents a multi-code situation in the downlink, in which case the base station 12 ₁ transmits 2 codes for each UE 14 ₁ . The trends shown in Figures 10 and 11 are similar to the observed performance trends for SIC-LSE shown in Figures 8 and 9. In the range of 1%-10% BER, the performance of SIC-LSE is equivalent to that of JD (that is, the difference is not more than 1dB), and this range is the working range of non-coded BER that people care about. When the power difference is 1-2dB (depending on different channels), the performance of SIC-LSE is comparable to that of full JD. As shown, its performance increases with increasing power difference between the two bursts.

Fig. 12 is similar to Fig. 10, except that there are only two burst groups in Fig. 12, and each group contains 4 bursts. As shown in Figure 12, in the range of 1%-10% BER, the performance of SIC-JD is comparable to that of JD (ie, the difference is not greater than 1dB).

的元素属于4个QPSK构象(格局)点之一，因此可以计算出4n_i个可能向量，这对于计算乘积 是必需的。该步要求每秒进行 $4\mathrm{\α}\·(\mathrm{SF}+W-1)\·\frac{\mathrm{Rate}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\Σ}}}{n}_i$ 百万次实运算(MROPS)。其中α＝4为执行一次复数乘法运算或乘法与累积(MAC)运算所进行的实运算次数；Rate为每秒进行的SIC-JD次数。由于已计算出上述4n_i个向量，

的计算要求每秒进行 $\frac{\mathrm{\α}}{2}\·N_s\·(\mathrm{SF}+W-1)\·\frac{\mathrm{Rate}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\Σ}}}{n}_i$ 百万次实运算。由于只进行复数加法运算，因此执行一次复数运算只需进行两次实运算，故上述公式采用系数

。由此，G-1级干扰消除的复杂度可由公式11表示。 $z=\mathrm{\α}(\mathrm{SF}+W-1)\·(4+\frac{N_s}{2})\·\frac{\mathrm{Rate}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\Σ}}}{n}_i$ 公式11The complexity of SIC-JD is lower than that of full JD. The reduction in complexity stems from the use of G JD-level dimensional matrices N _c ×n _i ·N _s (i=1 to G) instead of single-step JD-dimensional matrices N _c ×K·N _s . At the same time, since JD generally involves matrix inversion, and the complexity of inversion is proportional to the cube of the number of bursts, the overall complexity of multi-stage JD will be much lower than that of single-stage full JD. Moreover, the complexity of the SIC part has only a linear relationship with the number of bursts, so the advantage of SIC-JD in terms of complexity will not be significantly weakened. For example, the complexity of G-1 level interference cancellation can be derived as follows. Since the serial block of A _g ⁽ⁱ⁾ is the shifted version of the first block, and it is assumed that

The elements of belong to one of the 4 QPSK conformation (pattern) points, so 4n _i possible vectors can be calculated, which is useful for calculating the product is compulsory. This step is required to be performed every second $4\mathrm{\&alpha;}\&Center\; Dot;(\mathrm{SF}+W-1)\&Center\; Dot;\frac{\mathrm{<figure-callout\; id="10"\; label="Rate"\; filenames="A0180945200171.png,A0180945200221.png"\; state="\{\{state\}\}">Rate</figure-callout>}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\&Sigma;}}}{\mathrm{no}}_i$ Million real operations (MROPS). Wherein, α=4 is the number of real operations performed by performing a complex multiplication operation or a multiply and accumulate (MAC) operation; Rate is the number of times of SIC-JD performed per second. Since the above 4n _i vectors have been calculated,

calculations required to be performed per second $\frac{\mathrm{\&alpha;}}{2}\&Center\; Dot;N_{\mathrm{the\; s}}\&CenterDot;(\mathrm{SF}+W-1)\&Center\; Dot;\frac{\mathrm{<figure-callout\; id="10"\; label="Rate"\; filenames="A0180945200171.png,A0180945200221.png"\; state="\{\{state\}\}">Rate</figure-callout>}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\&Sigma;}}}{\mathrm{no}}_i$ Millions of actual operations. Since only complex addition is performed, only two real operations are required to perform one complex operation, so the above formula uses the coefficient

. Therefore, the complexity of G-1 level interference cancellation can be expressed by Equation 11. $z=\mathrm{\&alpha;}(\mathrm{SF}+W-1)\&Center\; Dot;(4+\frac{N_{\mathrm{the\; s}}}{2})\&CenterDot;\frac{\mathrm{<figure-callout\; id="10"\; label="Rate"\; filenames="A0180945200171.png,A0180945200221.png"\; state="\{\{state\}\}">Rate</figure-callout>}}{{10}^6}\underset{i=1}{\overset{G-1}{\mathrm{\&Sigma;}}}{\mathrm{no}}_i$ Formula 11

The complexity of soft-decision to hard-decision conversion is negligible.

There are several known methods to implement JD's matrix inversion. To illustrate its complexity, a very efficient approximate Cholesky factor algorithm is used, which involves negligible performance loss compared to the exact Cholesky factor algorithm. This algorithm can be used to solve group JD. See Table 1 for the complexity of SIC-JD and full JD in the 3GPP UTRA TDD system. Table 1 compares the complexity of various groups of different sizes. It can be seen that the advantage of SIC-JD over full JD in terms of complexity increases as K increases or the group size decreases. When the group size is 1, the complexity of SIC-LSE is linear with K, and when K=16, its complexity is 33% of full JD. Note: The maximum number of bursts in a UTRA TDD system is 16. When the exact Cholesky decomposition algorithm is used, the advantage of SIC-JD over full JD in terms of complexity will be more significant. Since the exact Cholesky decomposition algorithm has a stronger dependence on K, its complexity will be further reduced while reducing the dimension of JD by SIC-JD. total number of bursts The complexity of SIC-JD, expressed as a percentage of the single-step JD complexity of all K bursts Group K, each group size 1 (SIC-LSE) K/2 groups, each group size is 2 K/4 groups, each group size is 4 K/8 groups, each group size is 8 8 63% 67% 76% 100% 16 33% 36% 41% 57%

As shown in Table 1, the complexity of SIC-JD is on average lower than that of full JD when the code number and size become fully adaptive on a per-monitoring interval basis. On average, due to the difference in the grouping threshold, all bursts arriving at the receiver will not have the same power, so the group size will be smaller than the total number of bursts arriving. Alternatively, peak complexity may be reduced if the maximum allowed group size is hard-limited to be smaller than the maximum possible burst number. This approach results in some performance degradation when the bursts arriving at the receiver are of roughly equal power and the number of bursts exceeds the maximum allowed group size. To this end, SIC-JD provides a mechanism that achieves a balance between performance and peak complexity or required peak processing power.

Claims (25)

1. one kind is used for receiving the method for sharing a plurality of data-signals of frequency spectrum transmission by time slot at the code division multiple access tdd communication systems, and this method comprises:

Receive a composite signal by the shared frequency spectrum in this time slot;

A plurality of data-signals are grouped into a plurality of cohorts;

According to the part symbol response relevant this composite signal is carried out matched filtering with data-signal in one of cohort;

The data of all data-signals in group of joint-detection;

Part according to a group detects interference signal of data generation;

From composite signal, deduct the interference signal that is generated; And

Subtract by processing and signal measuring from other one group data.

2. method according to claim 1, wherein said joint-detection adopt the least squares prediction method to implement.

3. method according to claim 1, wherein said joint-detection adopt the least mean-square error estimation algorithm to implement.

4. one kind is used for receiving the method for sharing a plurality of data-signals of frequency spectrum transmission by a time slot at the tdd communication systems of employing code division multiple access, and it comprises:

(a) receive a composite signal as input signal by the shared frequency spectrum in this time slot;

(b) a plurality of data-signals are grouped into a plurality of cohorts, wherein at least one cohort comprises a plurality of data-signals;

(c) according to the part symbol response relevant this composite signal is carried out matched filtering with each data-signal in first group of the cohort;

(d) data of each data-signal in first group of the joint-detection;

(e) detect data according to first group part and generate an interference signal;

(f) from input signal, deduct the interference signal of generation, as the input signal of subsequent treatment;

(g) according to the part symbol response relevant this is subtracted signal and carry out matched filtering with follow-up one group data-signal in the cohort;

(h) data of each data-signal in follow-up one group of the joint-detection; And

(i) to each remaining set in cohort repeating step successively (e) to (h), wherein for each remaining set, subsequent group is as first group of this remaining set, this remaining set is as subsequent group.

5. method that is used for code division multiple access tdd communication systems receiver, wherein said method are used to receive by a time slot shares a plurality of data-signals that frequency spectrum sends, and this method comprises:

Receive a composite signal by the shared frequency spectrum in this time slot;

Estimate the received power level of each data-signal;

Received power level according to the partial data signal carries out the selectivity grouping to described a plurality of data-signals, and is divided into one group at least; And

Detect the data of data-signal in each group separately.

6. method according to claim 5, the estimation part of wherein said each data-signal received power level is according to the priori of receiver.

7. method according to claim 5, the estimation part of wherein said each data-signal received power level is according to the power stage of the training sequence relevant with each data-signal.

8. method according to claim 5 wherein uses one group of matched filter that the received power level of each data-signal is estimated, each filter all is complementary with a corresponding data signal code.

9. the data-signal that method according to claim 5, wherein said selective data signal packets mode will be in certain threshold range of power levels is divided into one group.

10. method according to claim 9, wherein said certain threshold power stage is 1 decibel.

11. method according to claim 9, wherein said certain adjustable threshold value is whole, in order to the receiver error rate that obtains to wish.

12. method according to claim 5, this method also comprises: all data-signals are forced to divide in a group, to surmount selectivity grouping step.

13. method according to claim 5, this method also comprises: each data-signal is forced to place group separately, to surmount selectivity grouping step.

14. a method that is used for receiver, the data signal data that this method can send the shared frequency spectrum of time slot in the code division multiple access tdd communication systems are used to adjust the balance between complexity and the performance when detecting, this method comprises:

Data-signal is divided into one group at least; Wherein, the data signal group number be increased, and, the data signal group number be reduced for improving performance for reducing complexity; And

Data in each group of joint-detection.

15. method according to claim 14, this method also comprises:

Determine the received power of each data-signal, wherein will divide into groups, all data-signals in same group are in certain threshold power stage data-signal; And when need reducing complexity, can increase this threshold value, and when needing to improve performance, can reduce this threshold value.

16. method according to claim 14, wherein in order to reduce complexity, each group all comprises a data-signal.

17. method according to claim 14, wherein in order to improve performance, having a group at least is single sets of signals.

18. a receiver that is used for the code division multiple access tdd communication systems, this system uses the multiple communication burst in the time slot to communicate, and described receiver comprises:

An antenna that is used to receive the radiofrequency signal that comprises multiple communication burst;

One is used for the demodulation radiofrequency signal to generate the demodulator of baseband signal;

A channel estimating apparatus that is used to estimate the pulse train channel response;

A serial interference elimination joint-detection (SIC-JD) device, it comprises:

One first is detected piece simultaneously, and this detection piece is used to detect the data that baseband signal comprised of the first group pulse string of being made up of a plurality of pulse trains;

One is used to generate the first group pulse string and disturbs first of estimated value to disturb the generation piece;

One is used for subtracting the subtracter that first group of interference from baseband signal; And

One second joint-detection piece, this detection piece are used for detecting the second group pulse string of being made up of a plurality of pulse trains and subtract the data that signal.

19. receiver according to claim 18, wherein said SIC-JD device also comprises:

A plurality of other joint-detection pieces, this detection piece is used to detect the data of all the other burst blocks of being made up of a plurality of pulse trains.

20. receiver according to claim 18, wherein said SIC-JD device also comprises:

One first matched filter, this filter is used to handle baseband signal, so that the symbol response of data-signal is complementary in first group; And

One second matched filter, this filter are used for handling to subtract signal, so that the symbol response of data-signal is complementary in second group.

21. receiver according to claim 18, wherein: one of first and second joint-detection piece is output as soft symbol, and the SIC-JD device also comprises first and second soft-decision-hard decision piece that is used for the output of first and second joint-detection piece is converted to hard symbol.

22. a device that is used for code division multiple access tdd communication systems receiver, this communication system use, and a plurality of communication burst communicate in the time slot, this device comprises:

An input unit, this configuration of cells are used for receiving and the relevant baseband signal of time slot received pulse string;

One first joint-detection piece, this detection piece are used to detect the data in first burst blocks baseband signal of being made up of the received pulse string.

First of an interference estimated value that is used to generate the first group pulse string disturbs and generates piece.

One is used for subtracting the subtracter that first group of interference from baseband signal; And

One second joint-detection piece, this detection piece are used to detect second burst blocks being made up of the received pulse string and subtract the data that in the signal.

23. device according to claim 22, this device also comprise other joint-detection piece that is used to detect other burst blocks data of being made up of a plurality of pulse trains.

24. device according to claim 22, this device also comprises:

One first matched filter, this filter is used to handle baseband signal, and the symbol response of received pulse string in first group is complementary; And

One second matched filter, this filter are used for handling to subtract signal, and the symbol response of received pulse string in second group is complementary.

25. device according to claim 22, one of the wherein said first and second joint-detection pieces is output as soft symbol; This device also comprises first and second a soft-decisions-hard decision piece that is used for the output of the first and second joint-detection pieces is converted to hard symbol.

Images